Google Gemini dubbed ‘high risk’ for kids and teens in new safety assessment

Common Sense Media, a kids-safety-focused nonprofit offering ratings and reviews of media and technology, released its risk assessment of Google’s Gemini AI products on Friday. While the organization found that Google’s AI clearly told kids it was a computer, not a friend — something that’s associated with helping drive delusional thinking and psychosis in emotionally vulnerable individuals — it did suggest that there was room for improvement across several other fronts.

Notably, Common Sense said that Gemini’s “Under 13” and “Teen Experience” tiers both appeared to be the adult versions of Gemini under the hood, with only some additional safety features added on top. The organization believes that for AI products to truly be safer for kids, they should be built with child safety in mind from the ground up.

For example, its analysis found that Gemini could still share “inappropriate and unsafe” material with children, which they may not be ready for, including information related to sex, drugs, alcohol, and other unsafe mental health advice.

The latter could be of particular concern to parents, as AI has reportedly played a role in some teen suicides in recent months. OpenAI is facing its first wrongful death lawsuit after a 16-year-old boy died by suicide after allegedly consulting with ChatGPT for months about his plans, having successfully bypassed the chatbot’s safety guardrails. Previously, the AI companion maker Character.AI was also sued over a teen user’s suicide.

In addition, the analysis comes as news leaks indicate that Apple is considering Gemini as the LLM (large language model) that will help to power its forthcoming AI-enabled Siri, due out next year. This could expose more teens to risks, unless Apple mitigates the safety concerns somehow.

Common Sense also said that Gemini’s products for kids and teens ignored how younger users needed different guidance and information than older ones. As a result, both were labeled as “High Risk” in the overall rating, despite the filters added for safety.

“Gemini gets some basics right, but it stumbles on the details,” Common Sense Media Senior Director of AI Programs Robbie Torney said in a statement about the new assessment viewed by TechCrunch. “An AI platform for kids should meet them where they are, not take a one-size-fits-all approach to kids at different stages of development. For AI to be safe and effective for kids, it must be designed with their needs and development in mind, not just a modified version of a product built for adults,” Torney added.

Google pushed back against the assessment, while noting that its safety features were improving.

The company told TechCrunch it has specific policies and safeguards in place for users under 18 to help prevent harmful outputs and that it red-teams and consults with outside experts to improve its protections. However, it also admitted that some of Gemini’s responses weren’t working as intended, so it added additional safeguards to address those concerns.

The company pointed out (as Common Sense had also noted) that it does have safeguards to prevent its models from engaging in conversations that could give the semblance of real relationships. Plus, Google suggested that Common Sense’s report seemed to have referenced features that weren’t available to users under 18, but it didn’t have access to the questions the organization used in its tests to be sure.

Common Sense Media has previously performed other assessments of AI services, including those from OpenAI, Perplexity, Claude, Meta AI, and more. It found that Meta AI and Character.AI were “unacceptable” — meaning the risk was severe, not just high. Perplexity was deemed high risk, ChatGPT was labeled “moderate,” and Claude (targeted at users 18 and up) was found to be a minimal risk.